Vision system for the automation of ovine carcass processing

Abstract

A vision system is presented for analysing the 3D profile of ovine carcasses for the belly rip down cut using a Microsoft Kinect. The system is an improvement over previous work, because it is significantly more robust against changes in the environment, while offering a simpler and more cost-effective setup. The integration of the vision system with a 6-axis robot manipulator is presented, leading to a fully automated vision-guided robotic system. Experimental trials involving 100 ovine carcasses were carried out, and a success rate of 99% was achieved. The vision system has a resolution of approximately 2 mm and experimental results show that it correctly identified the critical insertion point for the belly rip down cut. Challenges with the insertion point localisation are discussed and future work where improvements can be made are proposed.